Energy Economics 34 (2012) 125–131
Contents lists available at ScienceDirect
Energy Economics
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e n e c o
Does OPEC still exist as a cartel? An empirical investigation☆
Vincent Brémond a, Emmanuel Hache b, Valérie Mignon c,⁎
a
b
c
EconomiX-CNRS, University of Paris Ouest, and IFP Énergies Nouvelles, 1 et 4 avenue de Bois Préau, 92852 Rueil Malmaison, France
IFP Énergies Nouvelles, 1 et 4 avenue de Bois Préau, 92852 Rueil Malmaison, France
EconomiX-CNRS, University of Paris Ouest, and CEPII, Paris, France
a r t i c l e
i n f o
Article history:
Received 29 November 2010
Received in revised form 18 March 2011
Accepted 19 March 2011
Available online 30 March 2011
JEL classification:
C22
C23
L11
Q40
a b s t r a c t
The aim of this paper is to determine if OPEC acts as a cartel by testing whether the production decisions of the
different countries are coordinated and if they have an influence on oil prices. Relying on cointegration and
causality tests in both time series and panel settings, our findings show that the OPEC influence has evolved
through time, following the changes in the oil pricing system. While the influence of OPEC is found to be
important just after the counter-oil shock, our results show that OPEC is a price taker on the majority of the
considered sub-periods. Finally, by dividing OPEC between savers and spenders, we show that it acts as a
cartel mainly with a subgroup of its members.
© 2011 Elsevier B.V. All rights reserved.
Keywords:
Oil prices
Oil production
OPEC
Cartel
Cointegration
Causality
1. Introduction
Since the creation of the Organization of the Petroleum Exporting
Countries1 (OPEC) at the beginning of the 1960s, many authors have
focused on its role in the oil market and, notably, its capability to
influence oil prices in both the short and long run (see Dahl and Yücel
(1989) among others). At the same time, the oil pricing system
witnessed major transformations in the last 50 years, which can be
summarized by a shift from an administered pricing system to a
market related price since the middle of the 1980s. The posted price
period, with the different pricing systems in the physical market Gulf
price (Single Basis System), was implemented by the International Oil
Companies (IOC) at the beginning of the 20th century. According to
☆ We would like to thank two anonymous referees for helpful comments and
suggestions.
⁎ Corresponding author at: EconomiX-CNRS, University of Paris Ouest, 200 avenue de
la République, 92001 Nanterre Cedex, France. Tel.: + 33 1 40 97 58 60; fax + 33 1 40 97
77 84.
E-mail addresses: [email protected] (V. Brémond),
[email protected] (E. Hache), [email protected]
(V. Mignon).
1
OPEC was formed by five countries (Iran, Iraq, Kuwait, Saudi Arabia, and Venezuela)
in September 1960 in Baghdad.
0140-9883/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.eneco.2011.03.010
Yergin (1991) and Fattouh (2006), the aim of this pricing regime was
to lower the tax paid by the IOC to the host countries, leading to a very
low and stable official price whatever the market conditions. The
entry into the market of new large scale players, such as the Soviet
Union at the end of the 1950s, triggered a major change with a huge
surplus of production. In reaction, the so-called seven sisters cartel
decided to cut by 10% the posted prices in the market to safeguard
their market share. This factor can be considered as the key element
which triggered the creation of the OPEC in 1960. Nevertheless, it took
13 years for the Organization to claim its “market power” in the oil
sector. Since 1973, the oil market as well as the oil pricing regime has
thus experienced a period of continuous change.
Within this context, our aim is to investigate the dynamics of the
production behavior of countries belonging to the OPEC, as well as
non-member countries that are considered as major key players in the
oil market. More specifically, our aim is to determine if OPEC acts as a
cartel by testing whether the production decisions of the different
countries are coordinated and if they have an influence on oil prices.
A related question has been previously investigated by Dahl and
Yücel (1989) who test numerous theories about the OPEC behavior.
They show that OPEC is not a cartel, and that some countries behave in
a non-competitive way or with a target revenue goal. Considering
various sub-periods, Loderer (1985) shows that while the announcements of OPEC decisions do not affect prices in the 1974–1980 period,
126
V. Brémond et al. / Energy Economics 34 (2012) 125–131
140
120
100
80
60
40
20
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
0
Fig. 1. Oil price (US dollars per barrel, monthly frequency).
the alternative hypothesis that OPEC could act as a cartel is not
rejected during the beginning of the 1980s. Using cointegration and
Granger causality tests, Gülen (1996) obtains similar findings: OPEC
production Granger-causes oil prices during the 1982–1993 period,
whereas none causality is found for the previous periods investigated.
Testing different models (namely cartel models, competitive models,
target revenue models and property rights models), Griffin (1985)
puts forward that the partial market sharing is the best fitted model
for the OPEC countries, while non-OPEC countries are better
represented by the competitive model. Other studies consider that
OPEC is a divided cartel. Hnyilicza and Pindyck (1976) split OPEC in
two groups, namely “saver” and “spender” countries, the spender
countries being composed of members with an immediate need for
cash and rate of discount lower than the savers, and focuses on the
bargaining power between these two groups. Following the same
typology, Aperjis (1982) concludes that a conflict can exist between
OPEC members regarding their production decisions. There also exists
a set of studies concerned with target behavior models, such as Teece
(1982) and Adelman (1982) who modeled OPEC behavior according
to a target revenue model. Alhajji and Huettner (2000) conclude as
well that OPEC does not act as a cartel, and that the target revenue
model is not rejected for Algeria, Libya and Nigeria. Finally, one can
mention the works by Johany (1980) focusing on the impact of the
uncertainty about property rights, and MacAvoy (1982) showing that
political events and market fundamentals (a growing demand and
speculation) have played a key role in explaining the price dynamics
during the 1970s, more than the OPEC behavior itself.
This brief survey of the literature shows that no consensus exists
regarding the OPEC production behavior. Our aim is to contribute to
this literature by testing if OPEC acts as a cartel. To this end, we rely
on time series and panel (i) cointegration techniques to investigate
the existence of a long-term relationship between the production of
each member and that of the OPEC, and (ii) Granger causality tests
to apprehend the influence of OPEC production decisions on oil
prices.
The rest of the paper is organized as follows. In Section 2, we
briefly describe the major changes observed in the oil pricing regime
since the first oil shock, which also define the sub-periods of our
empirical study. Section 3 describes the data, and Section 4 provides
the results of cointegration and causality tests. Finally, Section 5
concludes the article.
2. Main changes in the oil pricing regime
In 1973, the oil market experienced a new way of pricing with the
introduction of the Government Selling Price (GSP) or Official Selling
Price (OSP). This price can be considered as the historical counter-part
of the Posted price implemented by the IOC few decades earlier except
that the prices were now determined by OPEC, with the Arabian Light
(34°, API) as the crude marker. OPEC was at that time the dominant
player in the market (with more than 50% of the market share).
Nevertheless, if this period was marked by the two world oil
shocks (1973–1974 and 1979–1980), with a sharp increase of the
prices on the market2—from 3.65 to 11.65 US dollars per barrel
between October and December 1973 in nominal terms, and from
around 14.00 US dollars in 1978 to 40.85 in November 1980—it also
had a significant positive impact on the level of production from
countries outside the Organization. Thus, during the 1973–1982
period, the non-OPEC countries’ production registered an increase
about 44.5%, with Mexico (+2.5 millions barrel per day), the United
Kingdom (+2.0 millions) and Norway (+0.5 million) representing
around 45% of this increase. It helps to develop the spot markets and
can explain the emergence of a dual system in terms of pricing with
the coexistence of an OPEC reference price (administered price) and a
market price.
This leads to the second sub-period of our study. The 1982–1986
period is characterized by an important decline of oil prices, given rise
to the oil counter-shock in 1986 with oil prices at less than
10 US dollars per barrel. It has also conducted to the implementation
of the quotas policy by the Organization in March 1982. From 1982 to
1986, the oil pricing system has experienced a transition period with
the administered OPEC price which lasts until 1985, a growing
influence of the price in the spot markets and the introduction of the
netback pricing system in 1986. This latter has been abandoned a few
months after due to the implementation of a new market share policy
from Saudi Arabia. The year 1986 can be considered as a milestone in
the oil markets with the introduction and the widespread of a new
system: the market related regime. It represents the first construction
phase of a complex structure on the market with the introduction
of a pricing system based on three reference prices: West Texas
2
See Fig. 1 which displays the evolution of monthly oil prices from January 1973 to
July 2009 (source: Datastream).
V. Brémond et al. / Energy Economics 34 (2012) 125–131
127
0,6
0,4
2009
2007
2005
2003
2001
1999
1997
1995
1993
1991
1989
1987
1985
1983
1981
1979
1977
1975
1973
0,2
Source: authors’ calculations from Datastream database.
Fig. 2. OPEC market share in world oil production.
Intermediate (WTI) for North America, Brent for Europe and South
American countries, and Oman–Dubai for crude oil sent to the East
Asia, by the national Mexican company PEMEX. It also represents a
sort of “golden youth” on the oil market including the import of the
classical tools of modern finance (swap, options) created by the
financial revolution, the so-called “financial big bang” of the early
1980s.
During 1986–1993, we observed a sharp increase of the liquidity
(in terms of financial contracts such as Light Sweet Crude Oil) in the
financial market, especially on the New York Mercantile Exchange
(NYMEX) and on the International Petroleum Exchange3 (IPE). This was
followed by a gradual financialization on the market during 1993–
2001. For example, the number of futures contracts in the NYMEX
increased by more than 40% during this period and at the same time
the oil-pricing regime experienced some controversy: squeeze,
decrease of the liquidity due to a marked decrease of the Brent and
WTI production which have introduced some doubts regarding the
efficiency of the price formation on the market. The choice of 1993 as a
break point may also be justified by the fact that, at that date, the OPEC
market shares started to stabilize around 40% after the high instability
observed during the 1980s with the consequences of the 2nd world oil
shock and the 1986 counter shock (Fig. 2).
After the introduction at the end of December 2000 of the law
modernizing raw materials markets—the Commodity Futures Modernization Act4 (CFMA)—two major changes have been registered.
From January 2001 to January 2009, we observed a sharp rise in
transaction volumes in the financial markets, and a context of high
volatility on the oil market. Between January and July 2008, oil prices
increased to almost 147 US dollars per barrel and collapsed a few
weeks later to under 35 US dollars per barrel. This context has left
many analysts and researchers puzzled by the underlying explanations for determination of prices and the influence of the noncommercial players in the market. The interaction between a physical
price based on geographical reference prices and a financial one based
on futures contracts with the underlying assumption of a growing
speculation factor seems to have changed drastically the market
conditions.
3
The IPE became the InterContinental Exchange (ICE) in 2001.
For more information, see the CFTC website at http://www.cftc.gov/lawandregulation/
index.htm.
4
This brief description of the main changes in the oil pricing system
leads us to consider five sub-periods in our empirical analysis: January
1973 to February 1982, March 1982 to April 1986, May 1986 to
February 1993, March 1993 to December 2000, and the period starting
in January 2001.
3. Data and unit root tests
We consider a sample of 15 countries including (i) 11 countries
belonging to the OPEC (Algeria, Indonesia, Iran, Iraq, Kuwait, Libya,
Nigeria, Qatar, Saudi Arabia, the United Arab Emirates, and Venezuela),
and (ii) 4 other non-OPEC countries (namely Mexico, Norway, the
United Kingdom, and Russia). These four countries have been chosen
because their productions account for around 25% of the whole
production (source: BP, 2010 statistics)—a high percentage that can be
illustrated by the fact that, as an example, the Russian production is
larger than the Saudi one. They are thus key actors of the current
market and historically major producers.5 Production and price series
are extracted from Datastream.
We use monthly data from January 1973 to July 2009. Note that for
some countries, production data are not available on the whole
period. More specifically, we consider Indonesian production until
May 2008, given that Indonesia leaves OPEC after that date. For
Mexico, Norway and UK, the analysis begins in 1982—due to the
establishment of a larger oil production policy than previously—while
the Russian analysis starts after its creation in 1991. Furthermore, we
exclude Ecuador and Gabon from the analysis because of their coming
and going into the Organization, as well as Angola because it is a “too
recent” member of the OPEC.
Recall that our aim is to test for a cooperative behavior between
OPEC members by investigating the link between the production of
one member country and the global production of the other OPEC
members. For each country i, we define the production of the other
member countries—which we call “rest of the cartel production”—as
the difference between the total OPEC production and the individual
production of country i.6
5
US and Chinese cases are not studied because these countries are oil importers.
Canada is also excluded from our analysis given that the Canadian production is
mostly led by the US market. Finally, we do not consider the former U.S.S.R countries—
such as Kazakhstan and Azerbaijan—due to data availability problems.
6
Note that in the appendix, the rest of cartel prod is denoted as “country2”.
128
V. Brémond et al. / Energy Economics 34 (2012) 125–131
Table 1
Individual and rest of the cartel productions integrated of order 1.
1973.01–1982.02
1982.03–1986.04
1986.05–1993.02
1993.03–2000.12
2001.01–2009.07
Algeria, Indonesia, Iran, Kuwait,
Libya, Nigeria, Qatar, Saudi
Arabia, Venezuela.
Algeria, Indonesia, Iraq,
Kuwait, Qatar, Venezuela,
Mexico, U.K.
Algeria, Indonesia,
Libya, Nigeria,
Qatar, U.K.
Algeria, Indonesia, Iraq, Libya, Nigeria,
Qatar, Saudi Arabia, U.A.E., Mexico,
Norway, U.K., Russia
Indonesia, Iran, Iraq, Kuwait, Libya, Nigeria,
Qatar, Saudi Arabia, U.A.E., Venezuela,
Mexico, Norway, U.K., Russia
The crude oil price is the UK Brent in US dollars. It is expressed in
real terms using the US CPI (extracted from Datastream) as the
deflator. Finally, note that production and price series are expressed in
logarithmic terms.
The first step is to determine the integration order of our series.
Given that our sample period is characterized by various oil pricing
regimes (see Section 2), we consider a test robust to structural breaks.
We rely on the Zivot and Andrews (1992) test. Under the null
hypothesis, there is a unit root without any exogenous structural
break, whereas the alternative hypothesis is the stationarity with a
break date determined endogenously. The conclusions of the test are
reported in Table A1 in Appendix A and show that the majority of the
considered series are I(1). Table 1 summarizes the results and lists the
countries for which both the individual production and the rest of the
cartel production are I(1).7
4. Cointegration and Granger causality tests
4.1. Time series analysis
In order to test for the existence of a long-run relationship
between the production of exporting countries and that of the OPEC,
we rely on the Engle and Granger (1987) test based on the null
hypothesis of no cointegration.
Results displayed in Table 28 show that, for the first three periods,
none of the large producing countries is cointegrated with the rest of
the Organization production. This result indicates that the individual
shares of production are not constant over time. Countries for which
cointegration is obtained do not account for more than 30% of the
OPEC production. Contrary to the first three periods, the 1993.03–
2000.12 sample is characterized by the presence of massive producers
(particularly Saudi Arabia) accounting for nearly 50% of the OPEC
market share. Moreover, the absence of cointegration is also rejected
for two countries belonging to our non-OPEC group, leading to the
question of a possible larger agreement between producer countries.
This sub-period could also be considered as some kind of “Golden
Age” in terms of coordination within the Organization. Finally, in the
last period (2001.01–2009.07), cointegration is observed for countries
representing about 65 to 70% of the OPEC production, but without any
non-OPEC producers. As for the previous period, Saudi Arabia belongs
to these cointegrated countries.
It is worth noting that during the first three periods characterized
by the absence of cointegration for the major oil producers, the OPEC
market share in world oil production was highly volatile and unstable
(see Fig. 2). On the contrary, this market share is quite stable since
1993. These findings may lead us to the conclusion that OPEC also fits
its production policy according to the non-OPEC supply and demand.
To investigate the direction of the link between production and
price series, we implement the Granger causality test. Due to the
presence of a unit root in the data, this test is applied to series in first
differences.
7
The countries that are not reported in this table correspond to the case where the
order of integration is not 1 for both production series. In addition, for some countries,
the number of dummies was very important on some sub-periods, such as Venezuela
on the 1982–1986 sub-period.
8
See Table A2 in Appendix A for the detailed results.
The first main conclusion that emerges from Table 39 is that,
except for Libya, we cannot reject the null hypothesis of no causality
for each OPEC member during the first controversial 1973.01–1982.02
sub-period (including two oil shocks). As a consequence, there is no
evidence that production policies are linked to price movements, a
result which is relevant with the fact that the price was fixed during
this period. The second sub-period is mainly characterized by the
rejection of the null hypothesis of no causality from price to
production for the OPEC as a whole, illustrating the fact that OPEC is
price taker. Whereas the links between price and production are
relatively weak in this period, the null of no causality from production
to price is rejected for major producers in the third period, with a
feedback effect observed for Saudi Arabia. The sub-period following
the counter shock is thus the only one for which the OPEC production
policy seems to have had an effect on price. It reflects the
abandonment of the “netback” pricing system from Saudi Arabia
and other OPEC countries which experienced the former system. For
the fourth period, the main finding is the existence of a relation
running from price to production for the OPEC. The last period is
probably one of the most interesting, because several producers
display a link from price to production in a context of increasing
prices. Once again, OPEC is price taker. The characteristics of these
countries are interesting as well. Actually, if we consider Saudi Arabia,
U.A.E., Kuwait, Libya and Iran, we can notice that, according to BP,
2010 Statistical Review, these countries have some of the largest
Reserves/Production (R/P) ratios (see also Table 4).
These results lead us to test if there is a split between countries
inside OPEC itself. We classify the member countries into two subgroups:
• “saver” group: Iran, Libya, Kuwait, Qatar, Saudi Arabia, U.A.E. and
Venezuela;
• “spender” group: Algeria, Indonesia, Iraq and Nigeria.
This separation aims at investigating if some countries have more
incentive to cheat than others. To explain this behavior, suppose that
cheating is induced by the short-run wish to earn money. Various
reasons may justify this behavior, depending notably on characteristics of the countries used to classify the members (see Table 4). The
first one relies on demographic facts: a country with a large
population and/or an increasing growth rate will have to “feed” its
population. The difficulty to meet population needs being higher in a
large population country, this provokes a requirement in money to
import goods in the short term more important than in a small
country. Assuming that this requirement in money is met by cheating
explains the presence of large population countries in our spender
group. Second, the type of crude oil may play an important role. It
seems reasonable to think that good quality crude (i.e. with the
lightest and lowest sulfur content) will be easily sold (and at a higher
price), whatever the price. Thus, the owner has a relative less
incentive to let oil in the ground, which involves faster reserves
depletion. Finally, the type of political regime matters. A democratic
regime supposes regular election dates and so, a short or mid-term
9
Detailed results are reported in Table A3 in Appendix A. Some countries that
belong to our sample are not displayed in Table 3. This is due to either data
unavailability, or different orders of integration (see details for the concerned
countries in Table A3 in Appendix A).
V. Brémond et al. / Energy Economics 34 (2012) 125–131
Table 2
Results of the Engle–Granger cointegration test.
Sub-periods
Cointegrated countries
1973.01–1982.02
1982.03–1986.04
1986.05–1993.02
1993.03–2000.12
2001.01-2009.07
Algeria, Kuwait, Nigeria, Qatar
Algeria, Indonesia, Mexico
Indonesia, Libya, Nigeria, Qatar
Nigeria, Qatar, U.A.E., Saudi Arabia, Mexico, Norway
Iran, Nigeria, Qatar, Saudi Arabiaa, U.A.E.a
a
Rejection of the null hypothesis of no cointegration at the 10% significance level.
Otherwise, the significance level is 5%.
view. Given that the population opinion obviously plays a crucial role
in winning votes and elections, accounting for it may justify the
“spending oil” behavior of countries to earn money. Assuming that
countries with democratic political regimes have a larger incentive to
spend oil revenues to please electors than economies with other
political regimes justifies that countries with democratic regimes
belong to our spender group.
While giving a global decomposition, it should be noticed that
these descriptive facts cannot always lead to a clear-cut distinction
between the two groups. Iran has a large population, in contrast to the
other subgroup members, but the large oil (and gas) reserves and its
strict political regime lead us to maintain the country in the saver
group. It is obvious that political regimes called “democratic”
(especially in Nigeria, where corruption is omnipresent) are not as
democratic as in the developed countries, but they are nonetheless
more democratic than the dynastic regimes.
To sum up, large population countries with small reserves that
have to take into account population needs (expressed through
elections) have more incentive to earn money and to cheat—and so to
belong to our spender group.
129
To give a few figures and illustrate the previous comments, Table 4
reports some descriptive characteristics for our two groups of
countries. The spender group is characterized by large population
countries (from 28 to 240 million), with a very valuable crude oil (but
with small reserves), and a (relatively) flexible political regime. The
case of Iraq deserves some comments. Its presence in the spender
group is due to three major wars, which involves that political regimes
have to rebuild the country. However, given the special environment
of Iraq—which is very unstable since at least 2003—it may be relevant
to remove this country from the analysis. Doing this reinforces our
decomposition according to the descriptive facts. Indeed, regarding
the spender group, Iraq has the lowest population and the highest R/P
ratio. Keeping Algeria, Nigeria and Indonesia, the spender group is
characterized by the following statistics: (i) countries with population
from 34 to 240 million, (ii) low R/P and (R/P)/population ratios
compared to countries belonging to the saver group, and (iii) a unique
type of political regime, namely democracy. The saver group—
characterized by a smaller discount rate than the spenders—is
composed by small population countries (between 830 000 and
66 million), with large R/P ratios, and relatively heavier crude oil.
These countries are less democratic than the other ones; the two
exceptions being Iran and Venezuela due to their specific political
regimes. Furthermore, despite a low R/P ratio, Qatar is included too
because of its high gas reserves, making oil business only as a
complement into the Qatari GDP. Let us now investigate the specific
properties of these two groups of countries through a panel data
analysis.
4.2. Panel data analysis
To complement our time series analysis, we now implement panel
cointegration tests by searching for a long-term relationship between
Table 3
Results of the Granger causality test.
1973.01–1982.02
1982.03–1986.04
1986.05–1993.02
1993.03–2000.12
2001.01–2009.07
Price-Prod
Libya
S. Arabia
UAE, Mexico, OPEC
Prod-Price
–
Venezuela, Mexico,
Qatar, OPECa
Algeria, Iraqa
Libya, Qatar, Algeria, Nigeria,
OPECa, S. Arabiaa
UAE
Iran, Kuwait, Libya, S. Arabia,
UAE, UK, Russia, OPEC
Venezuelaa
a
Rejection of the null hypothesis of no causality at the 10% significance level. Otherwise, the significance level is 5%. Price-prod: null of no causality from price to production, prod-price:
null of no causality from production to price.
Table 4
Descriptive facts.
Countries
Population
(in millions)
(CIA Factbook)
Pop. growth
rate (CIA
Factbook)
Rate of
unemployment
(CIA Factbook)
R/P
(BP stats)
(R/P)/pop
Current balance
(2008 and 2009)
(in $ billion)
Type of crude (crude quality,
sulfur content) (Energy
Intelligence Group)
Type of political regime
(CIA Factbook)
Algeria
Indonesia
Iraq
34.18
240.27
28.95
1.196
1.136
2.506
10.2%
7.7%
15.2%
18.5
11.8
(a)
0.54
0.05
N 3.45*
(34.45 and − 4.26)
(0.125 and 10.58)
(12.2 and − 19.9)
Republic
Republic
Republic (instable)
Nigeria
Iran
149.23
66.43
2
0.883
5%
11.8%
49.5
89.4
0.33
1.34
(39.36 and 10.01)
(23.99 and 2.653)
Kuwait
Libya
2.69
6.32
3.549 (b)
2.172
2.2%
30%
(a)
73.4
N 37.17*
11.61
(64.78 and 32.01)
(35.7 and 8.257)
Qatar
0.83
0.957
0.5%
54.7
65.90
(14.23 and 4.619)
28.69
1.848
11.7%
74.6
2.60
(133.5 and 26.5)
4.80
3.689
2.4%
(a)
N 20.83*
(22.31 and − 4.04)
26.81
1.508
7.9%
(a)
N 3.73*
(37.39 and 8.561)
Light, low sulfur content
Light, low sulfur content
Medium, low to medium
sulfur content
Medium, low sulfur content
Medium, medium to high
sulfur content
Medium, high sulfur content
Medium, low to medium
sulfur content
Low to high, medium sulfur
content
Medium, medium sulfur
content
Medium, low to medium
sulfur content
Heavy to oil sands, low to
extra high sulfur content
Saudi Arabia
U.A.E.
Venezuela
*Assumption based on a R/P ratio equal to 100 years. (a): more than 100 years, (b): reflects a return to pre-Gulf crisis immigration of expatriates.
Federal Republic
Islamic Republic
Constitutional Emirate
Authoritarian state, de facto
Emirate
Monarchy
Federation of Emirates
Federal Republic
130
V. Brémond et al. / Energy Economics 34 (2012) 125–131
Table 5
Panel cointegration tests (p-values).
Panel cointegration tests
All OPEC
1973.01–1982.02
1982.03–1986.04
1986.05–1993.02
1993.03–2000.12
2001.01–2009.07
Savers
1973.01–1982.02
1982.03–1986.04
1986.05–1993.02
1993.03–2000.12
2001.01–2009.07
Spenders
1973.01–1982.02
1982.03–1986.04
1986.05–1993.02
1993.03–2000.12
2001.01–2009.07
Group-mean panel cointegration tests
Panel v
Panel rho
Panel PP
Panel ADF
Group-mean rho
Group-mean PP
Group-mean ADF
0.1086
0.1720
0.2974
0.0275**
0***
[0.0032***]
0***
0.002***
0.2554
0.0135**
0***
[0.0002***]
0.0001***
0***
0.2700
0.0745*
0***
[0.0156**]
0***
0***
0.3439
0.1166
0***
[0.0162**]
0***
0.0002***
0***
0***
0***
[0.0002***]
0***
0***
0***
0***
0***
[0.0024***]
0***
0***
0***
0***
0***
[0.0029***]
0.3058
0.1923
0.2400
0.1025
0***
0***
0.0105**
0.2045
0***
0***
0.0017***
0***
0.2464
0.0010***
0.0002***
0.0012***
0***
0.3342
0.0012***
0***
0***
0.0066***
0.0008***
0***
0.0347**
0***
0.0002***
0.0017***
0***
0.0958*
0***
0.0002***
0.0002***
0.0001***
0.0022***
0***
0.3919
0.3942
0.1338
0***
[0***]
0***
0.1489
0.3983
0.3505
0***
[0***]
0.0053***
0.1329
0.3981
0.3976
0***
[0***]
0.0089***
0.0046***
0.3980
0.3915
0***
[0***]
0.0020***
0.0083***
0.0051***
0.0146**
0.0001***
[0.0001***]
0.0043***
0.0069***
0.0057***
0.0933*
0.0007***
[0.0002***]
0.0049***
0.0003***
0.0041***
0.1084
0.0009***
[0***]
*** (resp.**, *): rejection of the null hypothesis of no cointegration at the 1% (resp. 5%, 10%) significance level. Between squared brackets: results with excluding Iraq in the considered
sample.
the production of the various countries and the total OPEC production.
We consider the seven tests proposed by Pedroni (1999, 2004).10
These tests are based on the null hypothesis of no cointegration.
Among the 7 Pedroni's tests, 4 are based on the within dimension
(panel cointegration tests) and 3 on the between dimension (groupmean panel cointegration tests). Group-mean panel cointegration
statistics are more general in the sense that they allow for
heterogeneous coefficients under the alternative hypothesis: under
the alternative hypothesis, there exists a cointegration relationship,
and this relationship is not necessarily the same for each country.
Results from Pedroni's tests are reported in Table 5. As it is frequently
the case, these results are somewhat mixed. They can be summarized as
follows. Considering first the group-mean panel cointegration tests, the
null hypothesis of no cointegration is always rejected, meaning that a
long-term relationship exists between the production series, whatever
the panel and the sub-period considered. Turning now to the tests based
on the within dimension, the results are less clear-cut. Indeed, if the
homogeneity assumption of the cointegrating relationship between
countries is retained, our findings show that the null of no cointegration
is not rejected in the following cases: (i) the 1986.05–1993.02 subperiod for the three groups of countries, and (ii) the 1982.03–1986.04,
1986.05–1993.02 and 1993.02–2000.12 periods for the group of
spenders. To sum up, these findings tend to show that OPEC acts as a
cartel especially with the group of savers since there exists a long-term
relationship between the production of countries belonging to this
group and the total OPEC production. The influence of the OPEC seems
to be weakened on the 1986.05–1993.02 sub-period since no
cointegrating relationship exists on this sub-period, including the
case of the savers’ countries. This result is not surprising and illustrates
the decreasing influence of the OPEC following the oil counter-shock.
As previously mentioned, given the very unstable environment of
Iraq since 2003, we also consider all OPEC and spender groups
without including this country in our panels. Indeed, the presence of
Iraq may impact our findings given that oil production decisions are
expected to not follow the OPEC policy. Results in Table 5 show that
excluding Iraq does not change our results, since the null hypothesis of no
cointegration is rejected in all cases for the 2001.01–2009.07 sub-period.
In order to investigate the direction of the link between production
and price series, we now proceed to Granger-type causality tests.
Results based on first-differenced series are reported in Table 6. When
a causal link exists, it generally runs from price to production. Indeed,
a causality running from production to price is observed only in the
following cases: (i) on the 1986.05–1993.02 sub-period for the
complete panel and the group of savers, (ii) on the 1982.03–1986.04
sub-period for the group of spenders, and (iii) on the whole period for
the savers’ countries. These results tend to confirm the findings of the
cointegration tests since they put forward a higher influence of the
OPEC on the group of savers. They also highlight the growing role of
the OPEC during the 1982.03–1986.04 sub-period—with a causality
running from production to price in the group of spenders—
corresponding to the introduction of quotas and the development of
the spot market. As for the cointegration results, our findings in terms
of causality are robust to the exclusion of Iraq from our panels.
10
Panel unit root tests have been applied (see Table A4 in Appendix A) and show
that all production series are integrated of order 1, whether one considers the
complete panel or the two sub-samples. This confirms the findings obtained in the
time series framework.
*** (resp.**, *): rejection of the null hypothesis of no causality at the 1% (resp. 5%, 10%)
significance level. Price-prod: null of no causality from price to production, prod-price:
null of no causality from production to price. Between squared brackets: results with
excluding Iraq in the considered sample.
Table 6
Granger causality tests (p-values).
1973.01–
1982.02
1982.03–
1986.04
1986.05–
1993.02
1993.03–
2000.12
2001.07–
2009.07
All OPEC
Price-Prod
0.1869
0.0040***
0.0083***
0.0242**
Prod-Price
0.5534
0.1149
0.0157**
0.6285
0.0875*
[0.0010***]
0.8423
[0.6432]
Savers
Price-Prod
Prod-Price
0.2169
0.2189
0.0006***
0.1959
0.1157
0.0966*
0.0418**
0.5502
Spenders
Price-Prod
0.4736
0.1130
0.0332**
0.0822*
Prod-Price
0.9175
0.0409**
0.2196
0.7902
0.0001***
0.1057
0.1979
[0.2137]
0.6738
[0.5174]
V. Brémond et al. / Energy Economics 34 (2012) 125–131
5. Conclusion
In this paper, we analyze the evolution of the production behavior
of countries belonging to the OPEC, as well as four non-member
countries that are considered as key players in the oil market (Mexico,
Norway, Russia, and the UK). More specifically, we aim at determining
if OPEC acts as a cartel. To this end, we rely on time series and panel
cointegration and causality tests to investigate whether production
decisions of the different countries are coordinated and if they have an
influence on oil prices.
Our findings shows that the influence of OPEC has evolved through
time, following the changes in the oil pricing system registered on the
market. In particular, investigating the OPEC behavior on various subperiods, we find that, while OPEC's influence was strong in the period
that just follows the oil counter-shock, it acts as a price taker for the
majority of the considered sub-periods since 1973. Finally, by splitting
OPEC into two groups, the savers and spenders, we show that OPEC
may be viewed as a divided organization in the sense that it acts as a
cartel mainly with a subgroup of its members.
Appendix A. Supplementary data
Supplementary data to this article can be found online at
doi:10.1016/j.eneco.2011.03.010.
References
Adelman, M.A., 1982. OPEC as a cartel. In: Griffin, G.M., Teece, D.J. (Eds.), OPEC Behavior
and World Oil Prices. George Allen & Unwin, London.
131
Alhajji, A.F., Huettner, D., 2000. OPEC and other commodity cartels: a comparison.
Energy Policy 28, 1151–1164.
Aperjis, D., 1982. The Oil Market in the 1980s, OPEC Oil Policy and Economic
Development. Ballinger Publishing Company, Cambridge, MA.
BP, 2010. Statistical Review of World Energy. June.
Dahl, C., Yücel, M., 1989. Dynamic Modeling and testing of OPEC Behavior. Research
Paper 8917, Federal Reserve Bank of Dallas. .
Engle, R.F., Granger, C.W.J., 1987. Cointegration and error correction: representation,
estimation and testing. Econometrica 55, 251–276.
Fattouh, B., 2006. The origins and evolution of the current international oil pricing
system: a critical assessment. In: Mabro, R. (Ed.), Oil in the Twenty-First Century:
Issues, Challenges, and Opportunities. Oxford University Press, Oxford.
Griffin, J.M., 1985. OPEC behavior: a test of alternative hypotheses. The American
Economic Review 75 (5), 954–963.
Gülen, S.G., 1996. Is OPEC a cartel? Evidence from cointegration and causality tests. The
Energy Journal 17 (2), 43–57.
Hnyilicza, E., Pindyck, R.S., 1976. Pricing policies for a two-part exhaustible resource
cartel, the case of OPEC. European Economic Review 8, 139–154.
Johany, A.D., 1980. The Myth of the OPEC Cartel, the Role of Saudi Arabia. John Wiley
and Sons, New York.
Loderer, C., 1985. A test of the OPEC cartel hypothesis: 1974–1983. Journal of Finance 40
(3), 991–1006.
MacAvoy, P., 1982. Crude Oil Prices as Determined by OPEC and Market Fundamentals.
Ballinger Publishing Company, Cambridge, MA.
Pedroni, P., 1999. Critical values for cointegration tests in heterogeneous panels with
multiple regressors. Oxford Bulletin of Economics and Statistics S1 (61), 653–670.
Pedroni, P., 2004. Panel cointegration. Asymptotic and finite sample properties of
pooled time series tests with an application to the PPP hypothesis. Econometric
Theory 20, 597–625.
Teece, D., 1982. OPEC behaviour: an alternative view. In: Griffin, G.M., Teece, D.J. (Eds.),
OPEC Behavior and World Oil Prices. George Allen & Unwin, London.
Yergin, D., 1991. The Prize. The Quest for Oil, Money and Power. Simon & Schuster,
New York.
Zivot, E., Andrews, D., 1992. Further evidence of great crash, the oil price shock and unit
root hypothesis. Journal of Business and Economic Statistics 10, 251–270.